A genetic algorithm for multiple objective sequencing problems in mixed model assembly lines

Abstract

Presented in this paper is a new genetic algorithm to solve multiple objective sequencing problems in mixed model assembly lines. Mixed model assembly lines are a type of production line where a variety of product models similar in product characteristics are assembled. Such an assembly line is increasingly accepted in industry to cope with the recently observed trend of diversification of customer demands. While several objectives have been investigated for an efficient use of the production line, they are usually dealt with independently in existing researches. Relatively little attention has been paid to multiple objective problems. A new genetic algorithm is developed and is applied to the multiple objective problems addressed in this paper. The problem simultaneously considers three objectives found often in literature and meaningful in reality. The proposed genetic algorithm places emphasis on seeking for a set of diverse non-dominated solutions. Extensive computational experiments are carried out to demonstrate the superiority of the algorithm.Sequencing problems are important for an efficient use of mixed model assembly lines. There is a rich set of criteria on which to judge sequences of product models in terms of line utilization. We consider three practically important objectives: minimizing total utility work, keeping a constant rate of part usage and minimizing total setup cost. A considerate line manager would like to take into account all these factors. The multiple objective sequencing problem is described and its mathematical formulation is provided. A genetic algorithm is designed for finding near-Pareto or Pareto optimal solutions for the problem. A new genetic evaluation and selection mechanism, called Pareto stratum–niche cubicle, is proposed. The performance comparison of the proposed genetic algorithm with three existing genetic algorithms is made for various test-bed problems in terms of solution quality and diversity. The results reveal that the proposed genetic algorithm outperforms the existing genetic algorithms, especially for problems that are large and involve great variation in setup cost.